Evaluation of foods, drinks and diets in the Netherlands according to the degree of processing for nutritional quality, environmental impact and food costs

Evaluation of foods, drinks and diets in the Netherlands according to the degree of processing for nutritional quality, environmental impact and food costs

Evaluation of foods, drinks and diets

in the Netherlands according to the degree

of processing for nutritional quality,

environmental impact and food costs

Reina E Vellinga1*, Marieke van Bakel1, Sander Biesbroek2, Ido B Toxopeus1, Elias de Valk1, Anne Hollander1, Pieter van ’t Veer2 and Elisabeth H M Temme1

Abstract

Objective: This study investigates nutritional quality, environmental impact and costs of foods and drinks and their

consumption in daily diets according to the degree of processing across the Dutch population

Design: The NOVA classification was used to classify the degree of processing (processed foods (UPF) and

ultra-processed drinks (UPD)) Food consumption data were derived from the Dutch National Food Consumption Survey 2012–2016 Indicators assessed were nutritional quality (saturated fatty acids (SFA), sodium, mono and disaccharides (sugar), fibre and protein), environmental impact (greenhouse gas (GHG) emissions and blue water use) and food costs

Setting: The Netherlands.

Participants: Four thousand three hundred thirteen Dutch participants aged 1 to 79 years.

Results: Per 100 g, UPF were more energy-dense and less healthy than unprocessed or minimally processed foods

(MPF); UPF were associated with higher GHG emissions and lower blue water use, and were cheaper The energy and sugar content of UPD were similar to those of unprocessed or minimally processed drinks (MPD); associated with similar GHG emissions but blue water use was less, and they were also more expensive In the average Dutch diet, per

2000 kcal, ultra-processed foods and drinks (UPFD) covered 29% (456 g UPF and 437 g UPD) of daily consumption and 61% of energy intake UPFD consumption was higher among children than adults, especially for UPD UPFD con-sumption determined 45% of GHG emissions, 23% of blue water use and 39% of expenses for daily food consump-tion UPFD consumption contributed 54% to 72% to daily sodium, sugar and SFA intake

Conclusions: Compared with unprocessed or minimally processed foods and drinks, UPF and UPD were found to be

less healthy considering their high energy, SFA, sugar and sodium content However, UPF were associated higher GHG emissions and with less blue water use and food costs Therefore daily blue water use and food costs might increase

if UPF are replaced by those unprocessed or minimally processed As nutritional quality, environmental impacts and food costs relate differently to the NOVA classification, the classification is not directly applicable to identify win–win-wins of nutritional quality, environmental impact and costs of diets

© The Author(s) 2022 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which

permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line

to the material If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder To view a copy of this licence, visit http:// creat iveco mmons org/ licen ses/ by/4 0/ The Creative Commons Public Domain Dedication waiver (http:// creat iveco mmons org/ publi cdoma in/ zero/1 0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Open Access

*Correspondence: reina.vellinga@rivm.nl

1 Centre for Nutrition, Prevention and Health Services, National

Institute for Public Health and the Environment (RIVM), Antonie van

Leeuwenhoeklaan 9, Bilthoven 3721 MA, The Netherlands

Full list of author information is available at the end of the article

Providing healthy and sustainable diets is one of the

major challenges of this century Considering global

warming and the rise of nutrition-related

non-commu-nicable diseases (NCDs) [1], it is essential to identify,

understand, and influence key drivers that contribute to

unhealthy and unsustainable diets In the last few

dec-ades, the global nutritional transition is characterized

by a shift towards the consumption of ultra-processed

foods (UPF) at the expense of basic, unprocessed foods

[2 3] UPF are mostly or entirely created from substances

extracted from foods or derived from food constituents

and are transformed into unrecognizable, ready-to-eat

foods that contain additives and high amounts of energy,

sugar, fat and salt [4] In contrast, unprocessed or

mini-mally processed foods and drinks are those that are either

fresh or slightly altered to increase food safety,

accessibil-ity or palatabilaccessibil-ity

Food processing should be an integral part of a

sus-tainable food system [5 6] For instance, food processing

makes food safer, enables preservation of foods, helps to

overcome seasonal gaps, enables nose-to-tail

consump-tion and encourages reuse of materials [6] On the other

hand, food processing steps such as manufacturing,

packaging and distribution, contribute to GHG emissions

[7] Moreover, considerable amounts of energy, water and

packaging materials are used for food processing The

lat-ter significantly contributes to the plastic waste stream

entering marine ecosystems [7]

Processes and ingredients that are used to

manufac-ture UPF make them highly convenient for consumers

and highly profitable for manufacturers [4] Over the past

years, it has been argued that unhealthy foods are less

expensive compared with healthy foods while the price

gap between them is growing [8] Considering that food

prices are an important determinant of food choices and

nutritious diets, affordability of ultra-processed foods

seems inevitably linked to its consumption, which may

have implications for public health, health inequalities

and food security, among others [9]

Recent studies link UPF with adverse health outcomes

Higher availability or consumption of UPF is associated

with increased risk of overweight, obesity, cardiovascular

diseases (CVD), cancer and all-cause mortality [10–12]

In food-based dietary guidelines, several countries

rec-ommend reducing UPF consumption (for example, in

Brazil [13] and Canada [14]) or have set targets to reduce

UPF consumption (for example, by 20% in France by 2022

[15]) Existing literature on UPF has primarily focused on

nutrient profiles or health outcomes Less is known about

the association between UPF and environmental impact

or food costs

The NOVA classification is often used to categorize foods according to the degree of processing [4] It could potentially be used to distinguish nutritional quality, environmental impact and cost of diets If those indica-tors were consistently different in ultra-processed foods and drinks (UPFD) compared with unprocessed or mini-mally processed foods and drinks (MPFD), this would facilitate a win–win-win scenario for the transition towards a healthy and sustainable diet Therefore, this study examines the nutritional quality (via energy, satu-rated fatty acids (SFA), sodium, fibre, mono and disac-charides (sugar) and protein), environmental impact (via GHG emissions and blue water use) and food costs for UPFD compared with MPFD, as well as their consump-tion across a representative Dutch populaconsump-tion

Methods

Population and dietary data

Data for 4,313 Dutch children and adults aged 1 to

79  years were derived from the Dutch National Food Consumption Survey (DNFCS) 2012–2016 [16] Food consumption data was obtained using two 24-h non-consecutive dietary recalls and reported in Globodiet software (IARC©; former EPIC-Soft) [17] Background information such as date of birth, urbanisation level and educational level was collected by the market research agency who was responsible for the representativeness Information on body composition was gathered in dif-ferent ways depending on age: body weight and height

of 1–15-year-olds were measured, for 16–70-year-olds they were self-reported and body weight of < 70-year-olds was measured by a trained dietician Height was not measured for adults aged 71–79-years due to practical reasons.  Body Mass Index (BMI) was calculated as the average body weight (in kg) divided by average height (in m) squared (kg/m2) A full explanation and description of this survey are reported elsewhere [16] For the current study, participants were classified into subgroups based

on age (1–3, 4–8, 9–18, 19–30, 31–50 and 51–79  year-olds), weight status (underweight (BMI < 18.5  kg/m2), normal weight (BMI 18.5– < 25 kg/m2), overweight (BMI 25– < 30  kg/m2), and obese (BMI ≥ 30  kg/m2), level of education, and degree of urbanization The level of edu-cation was classified as low (primary eduedu-cation, lower vocational education, advanced elementary education), moderate (intermediate vocational education, higher sec-ondary education) or high (higher vocational education and university) The educational level concerned the par-ticipants’ highest completed educational level or, in the case of participants under the age of 19 years, of the head

of household The degree of urbanization was classified

as hardly urbanized (fewer than 1,000 addresses/km2),

moderately urbanized (1000–1500 addresses/km2) and

highly urbanized (1,500 or more addresses/km2) [16]

Degree of food processing

The NOVA food classification system was applied to

determine the degree of food processing [4] NOVA

cat-egorizes foods and drinks according to the nature, extent,

and purpose of the industrial processing they undergo

The classification distinguishes four categories:

unpro-cessed or minimally prounpro-cessed foods, prounpro-cessed culinary

ingredients, processed foods, and ultra-processed foods,

which are described in detail elsewhere [4] In the

cur-rent study, foods and drinks were classified into

sepa-rate categories Via facet descriptions from Globodiet,

all unique foods and drinks reported by participants

were identified and systematically categorized into one

of the four NOVA categories Ingredients of composite

dishes were individually reported The following facets

descriptions were used: conservation method (e.g fresh,

pasteurization, canned, frozen); production (e.g

indus-trial, ready-to-eat, fresh); medium (e.g in oil, in brine,

in syrup); salt content (e.g salted or not salted); sugar

content (e.g not sweetened or sweetened with sugar

and/or artificial sweeteners) and where appropriate

con-sistency/shape (e.g powder, liquid, sliced) Food groups

were based on Globodiet Food group-specific

categori-zation can be found in Supplemental Table S1 In short,

fresh or plain foods and drinks or slightly altered (dried,

frozen, steamed) were classified as unprocessed or

mini-mally processed foods (MPF) or drinks (MPD) such as

plain yoghurt, rice, coffee and tea Vegetable oils,

but-ter and other animal fats, and sugar were categorized as

processed culinary ingredients Fresh or slightly altered

foods combined with processed culinary ingredients

were classified as processed foods or drinks (e.g tuna in

oil, salted nuts) Foods and drinks that were either

ready-to-eat, industrially prepared, contained many additives,

emulsifiers and/or other comparable

formulations/ingre-dients were classified as ultra-processed (e.g fruity dairy

drinks, confectionery, margarine) All bread was

classi-fied as ultra-processed since most bread is industrially

prepared and contains food additives Alcoholic drinks

are not classified according to the NOVA classification

In the current study, wine, cider and beer were

classi-fied as processed as they are produced by fermentation

of unprocessed foods Other spirits and liquors (e.g gin

or whisky) were classified as ultra-processed A research

dietician cross-checked the classification and provided

expert judgement

Nutritional quality

Foods and drinks from the DNFCS 2012–2016 were

linked to food composition data of the Dutch Food

Composition Database (NEVO online version 2016/5.0)

in order to estimate daily intake of energy, SFA, sodium, mono and disaccharides (sugars), fibre and protein [18]

In addition to often assessed nutrients (e.g energy, SFA, sodium, sugar and fibre) that associate with UPFD con-sumption, protein is of importance since proteins plays

an important role in the transition towards a sustainable diet Mono and disaccharides were assessed since free or added sugar are not included in the Dutch food composi-tion table (NEVO-online version 2016/5.0)

Environmental impact

The environmental impacts of foods were evaluated for Greenhouse gas (GHG) emissions (in kg CO2-eq) and blue water use (in m3) Blue water use is also referred

to as irrigation water Data on environmental impact were derived from the Dutch Life Cycle Assessment (LCA) food database [19] In a previous study in which

we applied the LCA Food database we showed that the correlation between GHG emissions and other environ-mental indicators is generally high, except for blue water use [20] Therefore, this study examines, besides GHG emission, blue water use since this indicator focusses on other important foods which are ignored when solely focussing on GHG emissions In short, environmental impacts were based on LCA methodology, which quanti-fied the environmental impact through the foods’ entire life cycle LCAs had an attributional approach and hierar-chical perspective and were performed following the ISO

14040 and 14,044 guidelines A time horizon of 100 years was used, and GHG emissions were recalculated follow-ing Intergovernmental Panel on Climate Change (IPCC) guidelines (2006) [21] Economic allocation was applied when production processes led to more than one food product, except for milk, for which bio-physical alloca-tion was used The funcalloca-tional unit used was 1 kg of pre-pared food or drink on the plate, and converted to per

100 g The LCA food database provided primary data for

265 foods and drinks, which cover 75% of total amount of food intake These foods were previously selected based

on frequency of consumption in the DNFCS and varia-tion in types of food The environmental impact of foods and beverages for which primary data were not available but that were consumed in the DNFCS 2012–2016 were matched with similar foods The same methodology was applied in a previous study [20] In short, foods were matched by expert judgement of a panel of scientists and were based on similarities in types of food, produc-tion systems and ingredient composiproduc-tion For composite dishes, standardized recipes from the Dutch Food com-position table (NEVO-online version 2016/5.0) were used where available and if not available, recipes were based

on label information More detailed information on the

use of the database can be found elsewhere [19, 20]

Food costs

The Dutch food cost database was used to estimate food

costs A detailed description of the database can be found

elsewhere [22] Briefly, retail food prices (n = 902) of

the lowest, non-promotional price were collected from

a high segment supermarket (Albert Heijn) and a

dis-count supermarket (Lidl) during July and August 2017

in Amsterdam, the Netherlands Prices were adjusted

for the weight of packaging, preparation (shrinkage/

gain) and waste and expressed in € per 100 g edible

por-tion Eight hundred thirty-nine food prices were directly

linked to food composition data of the Dutch Food

Com-position Database (NEVO-online version 2016/5.0) and

covered 62% of the total amount of food intake [18]

Remaining foods were matched to similar foods based on

similarities in product, brand, (relative) price and

ingre-dient composition For composite dishes, standardized

recipes from the Dutch food composition table

(NEVO-online version 2016/5.0) were used

Data analysis

Descriptive statistics were applied to characterize the

nutritional and environmental indicators and costs for

foods and drinks (per 100 g) reported by DNFCS 2012–

2016, according to the degree of processing Primary

data was used to characterize environmental impact and

costs according to the degree of processing Notable

dif-ferences in characteristics between foods and drinks

per 100  g according to their degree of processing were

reported based on mean and 95%CI Daily average

con-sumption of UPFD, UPF and UPD was calculated over

two consumption days and expressed in weight (g) per

2000  kcal The outcomes were standardized in order to

assess the relative contribution of food intake according

to degree of processing towards the total dietary intake

Mann–Whitney U test or Kruskal–Wallis test for

non-normally distributed data and ANOVA for normal

dis-tributed data were applied to examine differences in

UPFD consumption across population subgroups

Nutri-tional quality (energy, SFA, sodium, sugar, fibre and

pro-tein), environmental impact (GHG emissions and blue

water use) and food costs for total diet and according to

degree of processing were calculated over two

consump-tion days and standardized to 2000  kcal per day and

were reported for total diet and according to degree of

processing Wilcoxon signed rank test for non-normally

distributed data and paired t-test for normal distributed

data were used to assess whether the nutritional

qual-ity, environmental impacts and food costs of the

con-sumption of culinary processed ingredients, processed

foods and drinks, and UPF and UPD differs from those

of unprocessed or minimally processed foods and drinks Descriptive statistics were reported as mean, 95% con-fidence interval (95%CI), 25th percentile, 50th percen-tile and 75th percentile (P25, P50, P75) Reported values were weighted for demographic properties, season, and combination of both consumption days (week or week-end) A sensitivity analysis was performed with alterna-tions made in the food classification for bread (processed instead of ultra-processed) The statistical analysis was performed using SAS software, version 9.4 (SAS Institute

Inc., Cary, NC, USA) A two-sided p-value of < 0.05 was

considered statistically significant

Results

Foods and drinks classified according to NOVA

Around half to two-thirds of the foods (54%) and drinks (62%) identified in DNFCS 2012–2016 were categorized

as ultra-processed foods (UPF) or drinks (UPD) (Fig. 1) Approximately a quarter of foods (25%) and one-third of drinks (31%) were classified as unprocessed or minimally processed foods (MPF) or drinks (MPD) In the food groups ‘Sugar, sweets and (savoury) snacks’ (98%), ‘Soft drinks’ (93%) ‘Grains and breads’ (76%), and ‘Fats and oils’ (71%), the majority of foods were classified as UPF or UPD The food groups ‘Eggs’ (0%), ‘Legumes’ (0%), ‘Veg-etables’ (1%), ‘Fish’ (8%), ‘Fruits’ (13%), ‘Tap water’ (0%) and ‘Fruit and vegetable juice’ (0%) contained a low or no share of UPF or UPD

Characteristics of ultra‑processed foods and drinks

UPF contained around double the amount of energy (313

vs 150 kcal/100 g (+ 109%)), triple the mono and disac-charides (16.1 vs 4.9 g/100 g (+ 229%)) and SFA (5.4 vs 1.9 g/100 g (+ 184%)), and four times the sodium (478 vs

126 mg/100 g (+ 279%)) compared with MPF (Table 1) UPF contained reasonably similar amounts of protein (7.1 vs 8.9 g/100 g) and fibre (2.3 vs 2.7 g/100 g) compared with MPF UPD had a similar energy (67 vs 75 kcal/100 g) and mono- and disaccharides (8.7 vs 7.3 g/100 g) content compared with MPD

UPF were associated with slightly higher GHG emis-sions (0.62 vs 0.55  kg CO2-eq/100  g (+ 12%)) but less usage of blue water (0.008 vs 0.033 m3/100  g (-97%)) compared with MPF Underlaying food groups showed

a large variation in average environmental impact, e.g GHG emissions were on average 0.19  kg CO2-eq/100  g for unprocessed or minimally processed vegetables while 2.75 kg CO2-eq/100 g for unprocessed or minimally pro-cessed meat UPD were associated with similar GHG emissions (0.11 vs 0.10  kg CO2-eq/100  g) but less blue water use (0.002 vs 0.008 m3/100  g (-75%)) than MPD UPF were almost half as expensive as MPF (€0.55 vs

€0.97/100 g (-43%)) UPD cost two times more (€0.37 vs

€0.15/100 g (+ 147%)) compared with MPD

Ultra‑processed foods and drinks in daily diets

The Dutch population consumed a daily absolute

aver-age of 3053 g (2126 kcal) of foods and drinks, of which

925  g UPFD (478  g UPF and 477  g UPD) The

abso-lute daily average UPFD consumption was 743  g for

1–3-year-olds, 1014  g for 4–8-year-olds, 1230  g for

9–13-year-olds, 1259  g for 14–18-year-olds, 1091  g for

19–30-year-olds, 959  g for 31–50-year-olds, 737  g for

51–70-year-olds and 617  g for 71–79-year-olds

Fig-ure 2shows the daily consumption of UPF and UPD by

age, in grams per 2000 kcal Per 2000 kcal, the daily

aver-age UPFD consumption was 893 g (456 g UPF and 437 g

UPD) and did not differ between men (889 g/2000 kcal)

and women (898  g/2000  kcal) (p > 0.05) (Table 2) Daily

UPFD consumption differs significantly between age

groups (p < 0.001) Children and teenagers up to 18 years

consumed, almost twice as much UPFD (approximately

1200 g/2000 kcal) compared with adults and older adults

aged 51 to 79  years (ranging between 632  g/2000  kcal

to 700  g/2000  kcal) Adults aged 19 to 30  years and

31 to 50  years consumed 962  g and 874  g UPFD per

2000  kcal, respectively Consumption of UPF ranged

from 438 to 485 g/2000 kcal for all age groups Children

and teenagers consumed more UPD (approximately

700 g/2000 kcal) than adults aged 19 to 50 years old (415

to 525  g/2000  kcal) and adults aged 51 to 79  years old

(ranging between 180 to 247 g/2000 kcal)

There were significant differences overall by subgroups

of education level and degree of urbanization (Table 2), ranging around 4–9% between the subgroups Partici-pants with a moderate education level (939 (95%CI 916, 962) g/2000 kcal) consumed 89 g more UPFD compared with higher educated participants (850 (95%CI 830, 871) g/2000  kcal) and 68  g more compared with lower edu-cated participants (871 (95%CI 838,903) g/2000  kcal)

(p < 0.001) Participants living in low urbanized areas

consumed 916 (95%CI 891, 942) g/2000 kcal UPFD, and consumed 40 or 20  g UPFD more than those living in highly or moderately urbanized areas, 876 (95%CI 856, 896) g/ 2000 kcal and 898 (95%CI 868, 928) g/ 2000 kcal

respectively (p < 0.01).

Nutritional quality, environmental impact and food costs

Although there was a statistically significant difference observed between UPF and MPF consumption, their con-sumption was more or less similar with 442 g/2000 kcal and 456  g/2000  kcal, respectively for UPF and MPF Energy intake from UPF was almost three times higher

at 1107  kcal (55%)compared with 372  kcal (19%) from

MPF (p < 0.001) Per 2000  kcal, UPF consumption

con-tributed most towards daily intake of sodium (1596 mg, 70%), fibre (11.1 g, 58%), SFA (15.3 g, 54%), protein (33 g, 44%) and mono and disaccharides (42 g, 40%) (Table 3) MPF consumption contributed less to daily nutrient intake, ranging between 7% (for sodium) and 37% (for fibre) The consumption of UPD (437  g/2000  kcal) was around three times lower than the consumption of MPD

Fig 1 Percentage of foods and drinks according to NOVA-categories for foods and drinks consumed in DNFCS 2012–2016 by food groups

Table

O 2

3 )/100 g

(1510  g/2000  kcal) (p < 0.001), contributed 6% to daily

energy intake and determined 25 g (24%) of daily sugar

intake

Compared with MPF, consumption of UPF contributes

more to GHG emissions (36% vs 30%) (p < 0.001) but less

to blue water use (19% vs 35%) (p < 0.001) per 2000 kcal

UPD determined approximately twice less GHG

emis-sions (7% vs 12%) (p < 0.001) and seven times less blue

water use (4% vs 27%) (p < 0.001) compared with MPD.

Dietary costs for UPF (€1.24/2000  kcal) and UPD

(€0.42/2000  kcal) consumption were lower compared

with costs of MPF (€1.32/2000 kcal) (p < 0.001) and MPD

(€0.63/2000 kcal) (p < 0.001) consumption.

Sensitivity analysis

In a sensitivity analysis, all bread was classified as

pro-cessed instead of ultra-propro-cessed The percentage UPF

in ‘Grains and breads’ decreased from 76 to 35% As a

result, the average fibre content of UPF decreased with

0.2 g fibre per 100 g (2.1 g fibre per 100 g) Daily average UPF consumption decreased from 456  g per 2000  kcal

to 336 g per 2000 kcal, resulting in an difference of 120 g (309  kcal) Obviously, UPF contributed less to daily intake of fibre (-6.3  g, -57%), protein (-12.6  g, -38%), sodium (-523 mg, -33%) and determined less GHG emis-sions (-0.14 kg CO2-eq, -8%), blue water use (-0.003 m3, -12%) and food costs (-€0.23, -19%)

Discussion

This study investigated nutritional quality, environ-mental impact and costs of foods, drinks and daily diets according to the degree of processing across the Dutch population Per 100  g, ultra-processed foods were on average energy-denser, less healthy, and associ-ated with higher GHG emissions but lower blue water use and were cheaper than unprocessed or minimally processed foods Per 100 g, ultra-processed drinks had

on average a similar energy and sugar content, similar

Fig 2 The daily average ultra-processed foods and drink consumption in grams per 2000 kilocalories for Dutch men and women aged 1 to

79 years according to different age groups

12% missings; < 0.05,

Table